Potential Differences in the Data (Deprecated)
Warning: The v2 Analytics API has been deprecated. See Ooyala IQ Analytics for more details on the new v3 Analytics API.There may be differences in the data in v3 Analytics (Ooyala IQ) when compared to data in v2 Analytics due to the following:
- Due to the re-processing of data to accommodate the multi-dimensional capabilities, the aggregation of the data may result in slightly different values than in v2 Analytics. We are also using a new robust architecture in Ooyala IQ which results in better error handling.
- Ooyala IQ uses Quova, a more accurate geo provider than in v2 Analytics.
- Ooyala IQ uses a more accurate device profiling provider than v2 Analytics, so the aggregation across devices and operating systems may differ. We are now using UADetector along with WURFL to optimize mobile and desktop coverage.
- The metrics are defined more accurately. Metrics with new definitions or bug
fixes in calculation include:
- Playthrough: v2 Analytics calculated playthrough on the server side based on the segments watched. There are 40 segments for each video. If the 10th segment got watched, Playthrough 25% was triggered; if the 30th segment got watched, Playthrough 75% was triggered. This behavior is problematic when people rewind and/or fast-forward. This calculation method did not always take deduplicating into account and did not report segments that users skipped, which at times led to playthrough 75% appearing higher than playthrough 25%, for example. The new analytics has a more accurate definition of playthrough, which is calculated on the client-side to indicate the furthest quartile point a user has reached within one viewing session. The new calculation deduplicates, reports for the segments that the user skipped, and has a more accurate measure of when the 100% mark is reached. This change in calculation provides more accurate metric data for playthrough. Data after August 1, 2014 is calculated using the new playthrough logic.
- Uniques: The calculation of unique metrics in v2 Analytics involved looking for cookies on the client side based on event timestamps. The new calculation of uniques is server-side and involves an algorithm called hyperloglog (HLL) and a guid that identifies browsers within a device. If the player does not see a guid a new guid is generated. This new calculation method reduces bugs, deduplicates, and is more accurate. You may see differences in your “unique” data, but the values should be similar, and any change is due to increased accuracy in “unique” calculation.
- Plays Requested: We have improved our calculation of this metric. You may see the number of plays requested from your HTML5 and SDK players increase after 11/11/2014 due to this implementation.
- Video Starts: We have improved our calculation of this metric. You may see the number of video starts from your HTML5 and SDK players decrease after 11/11/2014 due to this implementation.